#importing our data & libraries
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
df = pd.read_csv(r"/Users/dogayildiz/Desktop/modified_data.csv")
df.sort_values(by=['timestamp'])

#choosing 6 stocks
#print(df.columns.values)
#choose AKBNK,CCOLA,FROTO,PGSUS,THYAO,VESTL

#seeing if AKBNK has enough data
akbnk_timestampf = df.loc[df['AKBNK'].notna()]['timestamp'].iloc[0]
akbnk_timestampl = df.loc[df['AKBNK'].notna()]['timestamp'].iloc[-1]
print("akbnk start:",akbnk_timestampf)
print("akbnk finish:",akbnk_timestampl)

#seeing if CCOLA has enough data
ccola_timestampf = df.loc[df['CCOLA'].notna()]['timestamp'].iloc[0]
ccola_timestampl = df.loc[df['CCOLA'].notna()]['timestamp'].iloc[-1]
print("ccola start:",ccola_timestampf)
print("ccola finish:",ccola_timestampl)

#seeing if FROTO has enough data
froto_timestampf = df.loc[df['FROTO'].notna()]['timestamp'].iloc[0]
froto_timestampl = df.loc[df['FROTO'].notna()]['timestamp'].iloc[-1]
print("froto start:",froto_timestampf)
print("froto finish:",froto_timestampl)

#seeing if PGSUS has enough data
pgsus_timestampf = df.loc[df['PGSUS'].notna()]['timestamp'].iloc[0]
pgsus_timestampl = df.loc[df['PGSUS'].notna()]['timestamp'].iloc[-1]
print("pgsus start:",pgsus_timestampf)
print("pgsus finish:",pgsus_timestampl)

#seeing if THYAO has enough data
thyao_timestampf = df.loc[df['THYAO'].notna()]['timestamp'].iloc[0]
thyao_timestampl = df.loc[df['THYAO'].notna()]['timestamp'].iloc[-1]
print("thyao start:",thyao_timestampf)
print("thyao finish:",thyao_timestampl)

#seeing if VESTL has enough data
vestl_timestampf = df.loc[df['VESTL'].notna()]['timestamp'].iloc[0]
vestl_timestampl = df.loc[df['VESTL'].notna()]['timestamp'].iloc[-1]
print("vestl start:",vestl_timestampf)
print("vestl finish:",vestl_timestampl)

akbnk start: 2012-09-17 09:45:00
akbnk finish: 2019-07-23 18:00:00
ccola start: 2012-09-17 09:45:00
ccola finish: 2019-07-23 18:00:00
froto start: 2012-09-17 09:45:00
froto finish: 2019-07-23 18:00:00
pgsus start: 2013-04-26 09:30:00
pgsus finish: 2019-07-23 18:00:00
thyao start: 2012-09-17 09:45:00
thyao finish: 2019-07-23 18:00:00
vestl start: 2012-09-17 09:45:00
vestl finish: 2019-07-23 18:00:00

df['timestamp_2'] = pd.to_datetime(df['timestamp'])
df['year'] = df['timestamp_2'].dt.year
df['month'] = df['timestamp_2'].dt.month
df['day'] = df['timestamp_2'].dt.day
#chosen_indices = ['AKBNK', 'CCOLA', 'FROTO', 'PGSUS', 'THYAO', 'VESTL']

print(df)

                 timestamp    AEFES   AKBNK    AKSA  AKSEN   ALARK   ALBRK  \
0      2012-09-17 09:45:00  22.3978  5.2084  1.7102   3.87  1.4683  1.1356   
1      2012-09-17 10:00:00  22.3978  5.1938  1.7066   3.86  1.4574  1.1275   
2      2012-09-17 10:15:00  22.3978  5.2084  1.7102    NaN  1.4610  1.1356   
3      2012-09-17 10:30:00  22.3978  5.1938  1.7102   3.86  1.4537  1.1275   
4      2012-09-17 10:45:00  22.5649  5.2084  1.7102   3.87  1.4574  1.1356   
...                    ...      ...     ...     ...    ...     ...     ...   
50007  2019-07-23 17:00:00  20.4800  7.7300  9.1400   2.47  3.2300  1.2100   
50008  2019-07-23 17:15:00  20.5000  7.7200  9.1400   2.47  3.2200  1.2100   
50009  2019-07-23 17:30:00  20.5000  7.7400  9.1300   2.46  3.2300  1.2100   
50010  2019-07-23 17:45:00  20.4000  7.7000  9.1400   2.47  3.2400  1.2100   
50011  2019-07-23 18:00:00  20.4600  7.7000  9.1400   2.47  3.2300  1.2000   

        ANACM    ARCLK   ASELS  ...   VAKBN  VESTL   YATAS   YKBNK   YUNSA  \
0      1.0634   6.9909  2.9948  ...  3.8620   1.90  0.4172  2.5438  2.2619   
1      1.0634   6.9259  2.9948  ...  3.8529   1.90  0.4229  2.5266  2.2462   
2      1.0679   6.9909  2.9855  ...  3.8436   1.91  0.4229  2.5266  2.2566   
3      1.0679   6.9584  2.9855  ...  3.8529   1.91  0.4286  2.5324  2.2619   
4      1.0725   6.9909  2.9760  ...  3.8620   1.90  0.4286  2.5324  2.2619   
...       ...      ...     ...  ...     ...    ...     ...     ...     ...   
50007  2.8400  20.3000     NaN  ...  4.8600   9.98  5.3500  2.7500  4.2500   
50008  2.8400  20.3200     NaN  ...  4.8600   9.98  5.3400  2.7500  4.2400   
50009  2.8300  20.3400     NaN  ...  4.8600   9.96  5.3400  2.7600  4.2400   
50010  2.8200  20.3800     NaN  ...  4.8600   9.94  5.3400  2.7700  4.2400   
50011  2.8300  20.3200     NaN  ...  4.8500   9.93  5.3300  2.7700  4.2400   

        ZOREN         timestamp_2  year  month  day  
0      0.7789 2012-09-17 09:45:00  2012      9   17  
1      0.7789 2012-09-17 10:00:00  2012      9   17  
2      0.7789 2012-09-17 10:15:00  2012      9   17  
3      0.7860 2012-09-17 10:30:00  2012      9   17  
4      0.7789 2012-09-17 10:45:00  2012      9   17  
...       ...                 ...   ...    ...  ...  
50007     NaN 2019-07-23 17:00:00  2019      7   23  
50008     NaN 2019-07-23 17:15:00  2019      7   23  
50009     NaN 2019-07-23 17:30:00  2019      7   23  
50010     NaN 2019-07-23 17:45:00  2019      7   23  
50011     NaN 2019-07-23 18:00:00  2019      7   23  

[50012 rows x 65 columns]

import matplotlib.pyplot as plt

# List of chosen companies
chosen_companies = ['AKBNK','CCOLA', 'FROTO', 'PGSUS', 'THYAO', 'VESTL']

# Filter the data for the specified years (2017-2020)
start_year = 2017
end_year = 2020

for company in chosen_companies:
    # Filter the data for the company and the specified years
    filtered_company_data = df[df[f'{company}_outlier'] == False]
    company_data = filtered_company_data[
        (filtered_company_data['year'] >= start_year) & (filtered_company_data['year'] <= end_year)
    ]

    # Create a new figure and axis for each company
    fig, ax = plt.subplots(figsize=(12, 6))

    # Plot the company's stock price
    ax.plot(company_data['timestamp_2'], company_data[company])

    # Set axis labels and title
    ax.set_xlabel('Time')
    ax.set_ylabel('Stock Price')
    ax.set_title(f'{company} Stock Prices (2017-2020)')

    # Format the x-axis to show months and years
    import matplotlib.dates as mdates
    ax.xaxis.set_major_locator(mdates.MonthLocator(interval=1))  # Show every month
    ax.xaxis.set_major_formatter(mdates.DateFormatter('%b %Y'))  # Format as abbreviated month and year

    # Rotate x-axis labels for better readability
    plt.xticks(rotation=45)

    # Display the plot
    plt.tight_layout()
    plt.show()

# List of companies
companies = ['AKBNK', 'CCOLA', 'FROTO', 'PGSUS', 'THYAO', 'VESTL']

# Loop through companies and months
for company in companies:
    for year in range(2017, 2020):  # Assuming you want to cover 3 years
        for month in range(1, 13):  # For all months
            # Filter the data
            company_data = df[(df['year'] == year) & (df['month'] == month)][[company]]
            
            if not company_data.empty:
                # Create a boxplot
                plt.figure(figsize=(8, 6))
                sns.boxplot(data=company_data, orient="v", color='lightblue')
                plt.title(f'{company} Stock Prices in {month:02d}/{year}')
                plt.ylabel(f'{company} Stock Price')
                plt.show()

# Define a multiplier for the 3-sigma rule
sigma_multiplier = 3

# Create a new column to flag outliers
for company in companies:
    df[f'{company}_outlier'] = False  # Initialize the column as False

    for year in range(2017, 2020):  # Assuming you want to cover 3 years
        for month in range(1, 13):  # For all months
            # Filter the data for the specific company, year, and month
            data_subset = df[(df['year'] == year) & (df['month'] == month)][[company]]

            if not data_subset.empty:
                # Calculate the mean and standard deviation for the subset
                subset_mean = data_subset[company].mean()
                subset_std = data_subset[company].std()

                # Identify and flag outliers based on the 3-sigma rule
                lower_bound = subset_mean - sigma_multiplier * subset_std
                upper_bound = subset_mean + sigma_multiplier * subset_std
                # Define the condition for which rows will be considered as outliers
                outlier_condition = (
                    (df['year'] == year) & 
                    (df['month'] == month) & 
                    ((df[company] < lower_bound) | (df[company] > upper_bound))
                    )

                # Get the indices of rows that meet the outlier condition
                outlier_indices = df.loc[outlier_condition].index

                # Update the '_outlier' column to True for these indices
                df.loc[outlier_indices, f'{company}_outlier'] = True
                #df.loc[(df['year'] == year) & (df['month'] == month) & (
                    #(df[company] < lower_bound) | (df[company] > upper_bound)), f'{company}_outlier'] = True

# After running the loop, you will have a column for each company indicating whether each data point is an outlier.

# Extract the outliers
outliers_akbnk = df[df['AKBNK_outlier']]
outliers_ccola = df[df['CCOLA_outlier']]
outliers_froto = df[df['FROTO_outlier']]
outliers_pgsus = df[df['PGSUS_outlier']]
outliers_thyao = df[df['THYAO_outlier']]
outliers_vestl = df[df['VESTL_outlier']]

print("Outliers for AKBNK:")
print(outliers_akbnk[['timestamp', 'AKBNK']])
print("Outliers for CCOLA:")
print(outliers_ccola[['timestamp', 'CCOLA']])
print("Outliers for FROTO:")
print(outliers_froto[['timestamp', 'FROTO']])
print("Outliers for PGSUS:")
print(outliers_pgsus[['timestamp', 'PGSUS']])
print("Outliers for THYAO:")
print(outliers_thyao[['timestamp', 'THYAO']])
print("Outliers for VESTL:")
print(outliers_vestl[['timestamp', 'VESTL']])

Outliers for AKBNK:
                 timestamp   AKBNK
30420  2017-01-31 14:30:00  6.7643
30431  2017-01-31 17:15:00  6.7643
30432  2017-01-31 17:30:00  6.7803
30441  2017-02-01 11:15:00  6.6681
30442  2017-02-01 11:30:00  6.6521
31172  2017-03-06 15:45:00  7.4375
31173  2017-03-06 16:00:00  7.4375
31174  2017-03-06 16:15:00  7.4375
31175  2017-03-06 16:30:00  7.4375
31176  2017-03-06 16:45:00  7.4294
31177  2017-03-06 17:00:00  7.4294
31182  2017-03-07 09:45:00  7.4535
31183  2017-03-07 10:00:00  7.4294
31184  2017-03-07 10:15:00  7.4615
31185  2017-03-07 10:30:00  7.4375
31186  2017-03-07 10:45:00  7.4455
35793  2017-10-09 10:00:00  7.3946
35795  2017-10-09 10:30:00  7.4357
35796  2017-10-09 10:45:00  7.4439
35797  2017-10-09 11:00:00  7.4357
35798  2017-10-09 11:15:00  7.4192
35799  2017-10-09 11:30:00  7.4275
35800  2017-10-09 11:45:00  7.4192
35801  2017-10-09 12:00:00  7.4275
35802  2017-10-09 12:15:00  7.4357
35803  2017-10-09 12:30:00  7.4604
41023  2018-06-01 09:45:00  6.8976
44218  2018-11-01 09:45:00  5.6451
44219  2018-11-01 10:00:00  5.6365
44220  2018-11-01 10:15:00  5.5679
44221  2018-11-01 10:30:00  5.5507
44222  2018-11-01 10:45:00  5.5679
44223  2018-11-01 11:00:00  5.6193
44224  2018-11-01 11:15:00  5.6365
44225  2018-11-01 11:30:00  5.6536
44226  2018-11-01 11:45:00  5.7309
44227  2018-11-01 12:00:00  5.7223
44228  2018-11-01 12:15:00  5.7309
44229  2018-11-01 12:30:00  5.7394
44234  2018-11-01 14:15:00  5.7394
44239  2018-11-01 15:30:00  5.7309
44919  2018-12-03 09:45:00  6.6831
44920  2018-12-03 10:00:00  6.7003
44921  2018-12-03 10:15:00  6.6831
44922  2018-12-03 10:30:00  6.6917
44923  2018-12-03 10:45:00  6.6746
44924  2018-12-03 11:00:00  6.7089
44925  2018-12-03 11:15:00  6.7175
44926  2018-12-03 11:30:00  6.6746
44927  2018-12-03 11:45:00  6.6574
44928  2018-12-03 12:00:00  6.6402
44929  2018-12-03 12:15:00  6.6488
44930  2018-12-03 12:30:00  6.6574
44931  2018-12-03 12:45:00  6.6488
44932  2018-12-03 13:45:00  6.6317
Outliers for CCOLA:
                 timestamp    CCOLA
31099  2017-03-02 12:30:00  32.0245
35793  2017-10-09 10:00:00  33.1161
35794  2017-10-09 10:15:00  33.1723
35797  2017-10-09 11:00:00  33.2098
35798  2017-10-09 11:15:00  33.1910
35799  2017-10-09 11:30:00  33.2098
35801  2017-10-09 12:00:00  33.2098
35807  2017-10-09 14:00:00  33.1910
35808  2017-10-09 14:15:00  33.1910
35809  2017-10-09 14:30:00  33.1535
35810  2017-10-09 14:45:00  33.1910
35811  2017-10-09 15:00:00  33.1535
35812  2017-10-09 15:15:00  33.1348
35814  2017-10-09 15:45:00  33.1910
39634  2018-03-28 17:15:00  32.8724
39669  2018-03-30 09:45:00  37.1267
39934  2018-04-11 14:15:00  31.9354
39935  2018-04-11 14:30:00  31.8416
40489  2018-05-09 09:45:00  31.4856
44330  2018-11-06 14:30:00  29.4889
44510  2018-11-14 11:30:00  24.6826
47603  2019-04-01 09:45:00  29.0102
47604  2019-04-01 10:00:00  28.8187
47605  2019-04-01 10:15:00  28.9911
Outliers for FROTO:
                 timestamp    FROTO
35568  2017-09-27 16:45:00  37.3984
35570  2017-09-27 17:15:00  37.3984
35571  2017-09-27 17:30:00  37.3282
36519  2017-11-09 10:30:00  52.1848
36520  2017-11-09 10:45:00  51.8712
36521  2017-11-09 11:00:00  51.7816
36522  2017-11-09 11:15:00  51.7816
36536  2017-11-09 15:15:00  51.7816
36537  2017-11-09 15:30:00  51.7816
36538  2017-11-09 15:45:00  51.7368
36543  2017-11-09 17:00:00  51.7368
36544  2017-11-09 17:15:00  51.7816
36546  2017-11-09 17:45:00  51.8712
36547  2017-11-09 18:00:00  52.2295
36548  2017-11-10 09:45:00  52.2295
39607  2018-03-28 10:00:00  59.2622
39608  2018-03-28 10:15:00  59.3518
39609  2018-03-28 10:30:00  59.3069
41476  2018-06-22 17:45:00  64.9542
41478  2018-06-25 09:45:00  65.1402
41479  2018-06-25 10:00:00  65.4192
42881  2018-08-31 15:15:00  61.0951
42882  2018-08-31 15:30:00  61.0486
42884  2018-08-31 16:00:00  61.8855
42885  2018-08-31 16:15:00  63.9313
42886  2018-08-31 16:30:00  64.3033
42887  2018-08-31 16:45:00  64.9077
42888  2018-08-31 17:00:00  63.7453
42889  2018-08-31 17:15:00  63.2804
42890  2018-08-31 17:30:00  62.6294
42891  2018-08-31 17:45:00  62.2574
42892  2018-08-31 18:00:00  62.2574
Outliers for PGSUS:
                 timestamp  PGSUS
31030  2017-02-28 10:45:00  15.25
31031  2017-02-28 11:00:00  15.26
31032  2017-02-28 11:15:00  15.25
31033  2017-02-28 11:30:00  15.23
31034  2017-02-28 11:45:00  15.24
...                    ...    ...
48972  2019-05-31 17:00:00  34.82
48973  2019-05-31 17:15:00  34.78
48974  2019-05-31 17:30:00  35.10
48975  2019-05-31 17:45:00  35.22
48976  2019-05-31 18:00:00  35.20

[67 rows x 2 columns]
Outliers for THYAO:
                 timestamp  THYAO
31045  2017-02-28 15:15:00   5.46
31046  2017-02-28 15:30:00   5.45
31047  2017-02-28 15:45:00   5.42
31048  2017-02-28 16:00:00   5.42
31049  2017-02-28 16:15:00   5.44
31050  2017-02-28 16:30:00   5.46
31051  2017-02-28 16:45:00   5.45
31052  2017-02-28 17:00:00   5.46
31053  2017-02-28 17:15:00   5.46
31054  2017-02-28 17:30:00   5.45
31055  2017-02-28 17:45:00   5.45
31056  2017-02-28 18:00:00   5.46
42723  2018-08-17 10:15:00  14.65
42724  2018-08-17 10:30:00  14.35
42725  2018-08-17 10:45:00  14.48
42727  2018-08-17 11:15:00  14.67
42728  2018-08-17 11:30:00  14.62
42729  2018-08-17 11:45:00  14.57
42730  2018-08-17 12:00:00  14.60
42731  2018-08-17 12:15:00  14.52
42732  2018-08-17 12:30:00  14.58
42733  2018-08-17 12:45:00  14.57
42735  2018-08-17 13:45:00  14.59
47604  2019-04-01 10:00:00  12.62
47605  2019-04-01 10:15:00  12.60
47606  2019-04-01 10:30:00  12.70
Outliers for VESTL:
                 timestamp  VESTL
30208  2017-01-20 16:30:00   7.05
30209  2017-01-20 16:45:00   7.05
30210  2017-01-20 17:00:00   7.04
30212  2017-01-20 17:30:00   7.03
30213  2017-01-20 17:45:00   7.12
30214  2017-01-20 18:00:00   7.09
30215  2017-01-23 09:45:00   7.13
30216  2017-01-23 10:00:00   7.14
30217  2017-01-23 10:15:00   7.03
30218  2017-01-23 10:30:00   7.08
30219  2017-01-23 10:45:00   7.10
32419  2017-05-02 11:45:00   7.39
32420  2017-05-02 12:00:00   7.41
32421  2017-05-02 12:15:00   7.40
32422  2017-05-02 12:30:00   7.41
32423  2017-05-02 12:45:00   7.41
32425  2017-05-02 13:45:00   7.41
32426  2017-05-02 14:00:00   7.42
32427  2017-05-02 14:15:00   7.40
32431  2017-05-02 15:15:00   7.39
36313  2017-10-31 14:30:00   9.15
36314  2017-10-31 14:45:00   9.12
36315  2017-10-31 15:00:00   9.10
36316  2017-10-31 15:15:00   9.08
36317  2017-10-31 15:30:00   9.10
36318  2017-10-31 15:45:00   9.21
36319  2017-10-31 16:00:00   9.38
36320  2017-10-31 16:15:00   9.40
36321  2017-10-31 16:30:00   9.65
36322  2017-10-31 16:45:00   9.86
36323  2017-10-31 17:00:00   9.92
36324  2017-10-31 17:15:00   9.96
36325  2017-10-31 17:30:00   9.90
36326  2017-10-31 17:45:00   9.86
36327  2017-10-31 18:00:00   9.88
36338  2017-11-01 12:15:00  11.68
36343  2017-11-01 14:00:00  11.61
37685  2018-01-02 09:45:00   7.61
37686  2018-01-02 10:00:00   7.64
37687  2018-01-02 10:15:00   7.81
37688  2018-01-02 10:30:00   7.86
37689  2018-01-02 10:45:00   7.83
37690  2018-01-02 11:00:00   7.88
38923  2018-02-26 11:30:00  11.78
38924  2018-02-26 11:45:00  11.69
38925  2018-02-26 12:00:00  11.67
38926  2018-02-26 12:15:00  11.65
38927  2018-02-26 12:30:00  11.65
38928  2018-02-26 12:45:00  11.68
38929  2018-02-26 13:45:00  11.68
38930  2018-02-26 14:00:00  11.83
38931  2018-02-26 14:15:00  11.73
38932  2018-02-26 14:30:00  11.62
38944  2018-02-26 17:30:00  11.66
38945  2018-02-26 17:45:00  11.75
38946  2018-02-26 18:00:00  11.77
38947  2018-02-27 09:45:00  11.67
46902  2019-02-28 10:30:00  11.08
46903  2019-02-28 10:45:00  10.85

Insights for AKBANK Outliers

from IPython.display import display
Akbank_data = outliers_akbnk[['year', 'month', 'day', 'timestamp', 'AKBNK']]
display(Akbank_data)

	year	month	day	timestamp	AKBNK
30420	2017	1	31	2017-01-31 14:30:00	6.7643
30431	2017	1	31	2017-01-31 17:15:00	6.7643
30432	2017	1	31	2017-01-31 17:30:00	6.7803
30441	2017	2	1	2017-02-01 11:15:00	6.6681
30442	2017	2	1	2017-02-01 11:30:00	6.6521
31172	2017	3	6	2017-03-06 15:45:00	7.4375
31173	2017	3	6	2017-03-06 16:00:00	7.4375
31174	2017	3	6	2017-03-06 16:15:00	7.4375
31175	2017	3	6	2017-03-06 16:30:00	7.4375
31176	2017	3	6	2017-03-06 16:45:00	7.4294
31177	2017	3	6	2017-03-06 17:00:00	7.4294
31182	2017	3	7	2017-03-07 09:45:00	7.4535
31183	2017	3	7	2017-03-07 10:00:00	7.4294
31184	2017	3	7	2017-03-07 10:15:00	7.4615
31185	2017	3	7	2017-03-07 10:30:00	7.4375
31186	2017	3	7	2017-03-07 10:45:00	7.4455
35793	2017	10	9	2017-10-09 10:00:00	7.3946
35795	2017	10	9	2017-10-09 10:30:00	7.4357
35796	2017	10	9	2017-10-09 10:45:00	7.4439
35797	2017	10	9	2017-10-09 11:00:00	7.4357
35798	2017	10	9	2017-10-09 11:15:00	7.4192
35799	2017	10	9	2017-10-09 11:30:00	7.4275
35800	2017	10	9	2017-10-09 11:45:00	7.4192
35801	2017	10	9	2017-10-09 12:00:00	7.4275
35802	2017	10	9	2017-10-09 12:15:00	7.4357
35803	2017	10	9	2017-10-09 12:30:00	7.4604
41023	2018	6	1	2018-06-01 09:45:00	6.8976
44218	2018	11	1	2018-11-01 09:45:00	5.6451
44219	2018	11	1	2018-11-01 10:00:00	5.6365
44220	2018	11	1	2018-11-01 10:15:00	5.5679
44221	2018	11	1	2018-11-01 10:30:00	5.5507
44222	2018	11	1	2018-11-01 10:45:00	5.5679
44223	2018	11	1	2018-11-01 11:00:00	5.6193
44224	2018	11	1	2018-11-01 11:15:00	5.6365
44225	2018	11	1	2018-11-01 11:30:00	5.6536
44226	2018	11	1	2018-11-01 11:45:00	5.7309
44227	2018	11	1	2018-11-01 12:00:00	5.7223
44228	2018	11	1	2018-11-01 12:15:00	5.7309
44229	2018	11	1	2018-11-01 12:30:00	5.7394
44234	2018	11	1	2018-11-01 14:15:00	5.7394
44239	2018	11	1	2018-11-01 15:30:00	5.7309
44919	2018	12	3	2018-12-03 09:45:00	6.6831
44920	2018	12	3	2018-12-03 10:00:00	6.7003
44921	2018	12	3	2018-12-03 10:15:00	6.6831
44922	2018	12	3	2018-12-03 10:30:00	6.6917
44923	2018	12	3	2018-12-03 10:45:00	6.6746
44924	2018	12	3	2018-12-03 11:00:00	6.7089
44925	2018	12	3	2018-12-03 11:15:00	6.7175
44926	2018	12	3	2018-12-03 11:30:00	6.6746
44927	2018	12	3	2018-12-03 11:45:00	6.6574
44928	2018	12	3	2018-12-03 12:00:00	6.6402
44929	2018	12	3	2018-12-03 12:15:00	6.6488
44930	2018	12	3	2018-12-03 12:30:00	6.6574
44931	2018	12	3	2018-12-03 12:45:00	6.6488
44932	2018	12	3	2018-12-03 13:45:00	6.6317

# Filtering the data to include only dates between 2017 and 2019
filtered_df = df[(df['timestamp_2'] >= pd.to_datetime('2017-01-01')) & 
                 (df['timestamp_2'] <= pd.to_datetime('2019-12-31'))]

# Plotting each company's stock price trend with outliers for each month
company = 'AKBNK'
for year in range(2017, 2020):  # Years 2017 to 2019
    for month in range(1, 13):  # Months January to December
        monthly_data = filtered_df[(filtered_df['year'] == year) & (filtered_df['month'] == month)]
            
        if not monthly_data.empty:
            plt.figure(figsize=(10, 5))
            plt.plot(monthly_data['timestamp_2'], monthly_data[company], label=f'{company} Stock Price', linestyle='-')
                
            # Highlight outliers for the specific month
            monthly_outliers = monthly_data[monthly_data[f'{company}_outlier']]
            plt.scatter(monthly_outliers['timestamp_2'], monthly_outliers[company], color='red', label='Outliers', zorder=5)
                
            plt.title(f'{company} Stock Price Trend - {year}-{month:02d}')
            plt.xlabel('Date')
            plt.ylabel('Stock Price')
            plt.legend()
            plt.show()

We used 3-sigma rule to determine the outliers for each month. We listed the outliers for each company on the table above. Then, we ploted the graphs for each month to determine if there is a decrease or increase in trend.

First of all, we looked at Google Trends data using a search keyword "Akbank hisse" to observe if there were similar patterns.

For the first outlier, which is at the end of January, beginning of February of 2017. We can see that there is peak on the search volume for "Akbank hisse" on same days. We also searched for some news and we found that Central Bank of Republic of Türkiye announced an increase in the interest rates at Jan, 24. (https://www.ntv.com.tr/ekonomi/merkez-bankasindan-faiz-karari,-imZnKSSuU28UfK77sEoZg)

At 6-7 March of 2017, there is also an increase in stock prices. The Google Trends data is also corralated with that stock prices. We can see that search volume started to increase from 5 March to 10 March and then it started to decrease as in stock prices.

At Oct 9, 2017, there is a decrease in stock prices, which is another outlier. We also observed that google search volume hit the bottom at that particular day as we can see from the chart below.

Insights for COCA COLA Outliers

from IPython.display import display
ccola_data = outliers_ccola[['year', 'month', 'day', 'timestamp', 'CCOLA']]
display(ccola_data)

	year	month	day	timestamp	CCOLA
31099	2017	3	2	2017-03-02 12:30:00	32.0245
35793	2017	10	9	2017-10-09 10:00:00	33.1161
35794	2017	10	9	2017-10-09 10:15:00	33.1723
35797	2017	10	9	2017-10-09 11:00:00	33.2098
35798	2017	10	9	2017-10-09 11:15:00	33.1910
35799	2017	10	9	2017-10-09 11:30:00	33.2098
35801	2017	10	9	2017-10-09 12:00:00	33.2098
35807	2017	10	9	2017-10-09 14:00:00	33.1910
35808	2017	10	9	2017-10-09 14:15:00	33.1910
35809	2017	10	9	2017-10-09 14:30:00	33.1535
35810	2017	10	9	2017-10-09 14:45:00	33.1910
35811	2017	10	9	2017-10-09 15:00:00	33.1535
35812	2017	10	9	2017-10-09 15:15:00	33.1348
35814	2017	10	9	2017-10-09 15:45:00	33.1910
39634	2018	3	28	2018-03-28 17:15:00	32.8724
39669	2018	3	30	2018-03-30 09:45:00	37.1267
39934	2018	4	11	2018-04-11 14:15:00	31.9354
39935	2018	4	11	2018-04-11 14:30:00	31.8416
40489	2018	5	9	2018-05-09 09:45:00	31.4856
44330	2018	11	6	2018-11-06 14:30:00	29.4889
44510	2018	11	14	2018-11-14 11:30:00	24.6826
47603	2019	4	1	2019-04-01 09:45:00	29.0102
47604	2019	4	1	2019-04-01 10:00:00	28.8187
47605	2019	4	1	2019-04-01 10:15:00	28.9911

# Plotting each company's stock price trend with outliers for each month
company = 'CCOLA'
for year in range(2017, 2020):  # Years 2017 to 2019
    for month in range(1, 13):  # Months January to December
        monthly_data = filtered_df[(filtered_df['year'] == year) & (filtered_df['month'] == month)]
            
        if not monthly_data.empty:
            plt.figure(figsize=(10, 5))
            plt.plot(monthly_data['timestamp_2'], monthly_data[company], label=f'{company} Stock Price', linestyle='-')
                
            # Highlight outliers for the specific month
            monthly_outliers = monthly_data[monthly_data[f'{company}_outlier']]
            plt.scatter(monthly_outliers['timestamp_2'], monthly_outliers[company], color='red', label='Outliers', zorder=5)
                
            plt.title(f'{company} Stock Price Trend - {year}-{month:02d}')
            plt.xlabel('Date')
            plt.ylabel('Stock Price')
            plt.legend()
            plt.show()

There is a decrease in stock prices on May 9, 2018, as well as a sudden decrease in Google Trends search keyword “coca cola hisse”.

On 14 November 2018, Morgan Stanley Capital International Turkey removed the stocks of six companies from the Turkish index including CCOLA. (https://tr.investing.com/news/world-news/borsazayf-kuresel-risk-istah-ve-baz-turk-sirket-hisselerinin-msci-endeksinden-ckarlmasyla-bist100-basklanabilir-541331, https://tr.investing.com/news/stock-market-news/msci-turkiye-endeksinden-alt-hisseyi-ckard-541315) This might be a possible reason behind the fall of stock prices on the same date. Also, there is a incline on Google Trends.

On April 1, 2019, a decline in CCOLA stock prices can be seen. This decline is not reflected on Google Search Trends, however, we found some news on the same exact date indicating that Coca Cola is entering the energy drink market, releasing the first energy drink under Coca-Cola brand. (https://www.warc.com/newsandopinion/news/coca-cola-launches-its-first-branded-energy-drink/en-gb/41877) This might be one possible reason of the decline in its stock prices.

Insights for FORD OTOMOTİV Outliers

from IPython.display import display
froto_data = outliers_froto[['year', 'month', 'day', 'timestamp', 'FROTO']]
display(froto_data)

	year	month	day	timestamp	FROTO
35568	2017	9	27	2017-09-27 16:45:00	37.3984
35570	2017	9	27	2017-09-27 17:15:00	37.3984
35571	2017	9	27	2017-09-27 17:30:00	37.3282
36519	2017	11	9	2017-11-09 10:30:00	52.1848
36520	2017	11	9	2017-11-09 10:45:00	51.8712
36521	2017	11	9	2017-11-09 11:00:00	51.7816
36522	2017	11	9	2017-11-09 11:15:00	51.7816
36536	2017	11	9	2017-11-09 15:15:00	51.7816
36537	2017	11	9	2017-11-09 15:30:00	51.7816
36538	2017	11	9	2017-11-09 15:45:00	51.7368
36543	2017	11	9	2017-11-09 17:00:00	51.7368
36544	2017	11	9	2017-11-09 17:15:00	51.7816
36546	2017	11	9	2017-11-09 17:45:00	51.8712
36547	2017	11	9	2017-11-09 18:00:00	52.2295
36548	2017	11	10	2017-11-10 09:45:00	52.2295
39607	2018	3	28	2018-03-28 10:00:00	59.2622
39608	2018	3	28	2018-03-28 10:15:00	59.3518
39609	2018	3	28	2018-03-28 10:30:00	59.3069
41476	2018	6	22	2018-06-22 17:45:00	64.9542
41478	2018	6	25	2018-06-25 09:45:00	65.1402
41479	2018	6	25	2018-06-25 10:00:00	65.4192
42881	2018	8	31	2018-08-31 15:15:00	61.0951
42882	2018	8	31	2018-08-31 15:30:00	61.0486
42884	2018	8	31	2018-08-31 16:00:00	61.8855
42885	2018	8	31	2018-08-31 16:15:00	63.9313
42886	2018	8	31	2018-08-31 16:30:00	64.3033
42887	2018	8	31	2018-08-31 16:45:00	64.9077
42888	2018	8	31	2018-08-31 17:00:00	63.7453
42889	2018	8	31	2018-08-31 17:15:00	63.2804
42890	2018	8	31	2018-08-31 17:30:00	62.6294
42891	2018	8	31	2018-08-31 17:45:00	62.2574
42892	2018	8	31	2018-08-31 18:00:00	62.2574

# Plotting each company's stock price trend with outliers for each month
company = 'FROTO'
for year in range(2017, 2020):  # Years 2017 to 2019
    for month in range(1, 13):  # Months January to December
        monthly_data = filtered_df[(filtered_df['year'] == year) & (filtered_df['month'] == month)]
            
        if not monthly_data.empty:
            plt.figure(figsize=(10, 5))
            plt.plot(monthly_data['timestamp_2'], monthly_data[company], label=f'{company} Stock Price', linestyle='-')
                
            # Highlight outliers for the specific month
            monthly_outliers = monthly_data[monthly_data[f'{company}_outlier']]
            plt.scatter(monthly_outliers['timestamp_2'], monthly_outliers[company], color='red', label='Outliers', zorder=5)
                
            plt.title(f'{company} Stock Price Trend - {year}-{month:02d}')
            plt.xlabel('Date')
            plt.ylabel('Stock Price')
            plt.legend()
            plt.show()

There is a significant decrease at the end of September 2017. When we look at the google trends data, search volumes of "ford otosan bist" and "IST:FROTO" also decreased compared to the rest of the month.

After that decrease, the stock prices started to increase. At 9 November 2017, there is a peak, which is also an outlier data point. According to the news dated October 30, 2017, Ford Otosan's Q3 net profit increased by 98.9% to 345.9 million TL. (https://bigpara.hurriyet.com.tr/haberler/ekonomi-haberleri/ford-otosanin-k-ri-yuzde-989-artti_ID1441618/) After that news, the stock prices increased as we can see above.

Similar to the previous increase, we can attribute the increase in March to the news in this field. The total net profit of 8 companies operating in the automotive sector in Borsa Istanbul increased by 31.7 percent compared to the previous year, reaching 3.3 billion lira. (https://www.aa.com.tr/tr/ekonomi/borsada-otomotiv-sirketlerinin-toplam-kari-3-milyar-lirayi-asti/1092310) That news dated March 18, can lead to an increase in the stock prices at the following days.

On June 20, 2018, the merger of Volkswagen and Ford was officially announced. It was announced that the work would proceed by sharing the technology, infrastructure and development costs of large projects. (https://www.donanimhaber.com/Volkswagen-ve-Ford-gelecek-icin-guclerini-birlestirdi--100777) We can say that this news also increased share prices. This increase peaked on 22nd and 25th of the June.

Insights for PEGASUS Outliers

from IPython.display import display
pgsus_data = outliers_pgsus[['year', 'month', 'day', 'timestamp', 'PGSUS']]
pd.set_option('display.max_rows', None)
display(pgsus_data)

	year	month	day	timestamp	PGSUS
31030	2017	2	28	2017-02-28 10:45:00	15.25
31031	2017	2	28	2017-02-28 11:00:00	15.26
31032	2017	2	28	2017-02-28 11:15:00	15.25
31033	2017	2	28	2017-02-28 11:30:00	15.23
31034	2017	2	28	2017-02-28 11:45:00	15.24
31035	2017	2	28	2017-02-28 12:00:00	15.26
31036	2017	2	28	2017-02-28 12:15:00	15.24
31037	2017	2	28	2017-02-28 12:30:00	15.26
31040	2017	2	28	2017-02-28 14:00:00	15.26
31041	2017	2	28	2017-02-28 14:15:00	15.26
31042	2017	2	28	2017-02-28 14:30:00	15.24
31043	2017	2	28	2017-02-28 14:45:00	15.23
31044	2017	2	28	2017-02-28 15:00:00	15.22
31045	2017	2	28	2017-02-28 15:15:00	15.16
31046	2017	2	28	2017-02-28 15:30:00	15.10
31047	2017	2	28	2017-02-28 15:45:00	15.06
31048	2017	2	28	2017-02-28 16:00:00	15.12
31049	2017	2	28	2017-02-28 16:15:00	15.16
31050	2017	2	28	2017-02-28 16:30:00	15.19
31051	2017	2	28	2017-02-28 16:45:00	15.21
31052	2017	2	28	2017-02-28 17:00:00	15.18
31053	2017	2	28	2017-02-28 17:15:00	15.18
31054	2017	2	28	2017-02-28 17:30:00	15.14
31055	2017	2	28	2017-02-28 17:45:00	15.15
31056	2017	2	28	2017-02-28 18:00:00	15.16
40863	2018	5	24	2018-05-24 17:30:00	24.16
40864	2018	5	24	2018-05-24 17:45:00	23.86
40865	2018	5	24	2018-05-24 18:00:00	23.86
40866	2018	5	25	2018-05-25 09:45:00	24.04
40867	2018	5	25	2018-05-25 10:00:00	23.86
40868	2018	5	25	2018-05-25 10:15:00	24.06
40869	2018	5	25	2018-05-25 10:30:00	24.10
40870	2018	5	25	2018-05-25 10:45:00	24.10
40871	2018	5	25	2018-05-25 11:00:00	23.96
40872	2018	5	25	2018-05-25 11:15:00	24.04
40873	2018	5	25	2018-05-25 11:30:00	24.08
40874	2018	5	25	2018-05-25 11:45:00	24.16
40875	2018	5	25	2018-05-25 12:00:00	24.04
40876	2018	5	25	2018-05-25 12:15:00	23.88
40877	2018	5	25	2018-05-25 12:30:00	23.82
40878	2018	5	25	2018-05-25 12:45:00	23.72
40880	2018	5	25	2018-05-25 13:45:00	23.72
40881	2018	5	25	2018-05-25 14:00:00	23.76
40882	2018	5	25	2018-05-25 14:15:00	23.94
40883	2018	5	25	2018-05-25 14:30:00	24.02
42893	2018	9	3	2018-09-03 09:45:00	23.80
42894	2018	9	3	2018-09-03 10:00:00	24.38
42895	2018	9	3	2018-09-03 10:15:00	24.46
47604	2019	4	1	2019-04-01 10:00:00	26.88
47605	2019	4	1	2019-04-01 10:15:00	26.70
47606	2019	4	1	2019-04-01 10:30:00	26.84
47607	2019	4	1	2019-04-01 10:45:00	27.04
47608	2019	4	1	2019-04-01 11:00:00	27.00
48963	2019	5	31	2019-05-31 14:45:00	34.50
48964	2019	5	31	2019-05-31 15:00:00	34.92
48965	2019	5	31	2019-05-31 15:15:00	35.10
48966	2019	5	31	2019-05-31 15:30:00	35.20
48967	2019	5	31	2019-05-31 15:45:00	35.02
48968	2019	5	31	2019-05-31 16:00:00	34.98
48969	2019	5	31	2019-05-31 16:15:00	34.84
48970	2019	5	31	2019-05-31 16:30:00	34.82
48971	2019	5	31	2019-05-31 16:45:00	34.88
48972	2019	5	31	2019-05-31 17:00:00	34.82
48973	2019	5	31	2019-05-31 17:15:00	34.78
48974	2019	5	31	2019-05-31 17:30:00	35.10
48975	2019	5	31	2019-05-31 17:45:00	35.22
48976	2019	5	31	2019-05-31 18:00:00	35.20

# Plotting each company's stock price trend with outliers for each month
company = 'PGSUS'
for year in range(2017, 2020):  # Years 2017 to 2019
    for month in range(1, 13):  # Months January to December
        monthly_data = filtered_df[(filtered_df['year'] == year) & (filtered_df['month'] == month)]
            
        if not monthly_data.empty:
            plt.figure(figsize=(10, 5))
            plt.plot(monthly_data['timestamp_2'], monthly_data[company], label=f'{company} Stock Price', linestyle='-')
                
            # Highlight outliers for the specific month
            monthly_outliers = monthly_data[monthly_data[f'{company}_outlier']]
            plt.scatter(monthly_outliers['timestamp_2'], monthly_outliers[company], color='red', label='Outliers', zorder=5)
                
            plt.title(f'{company} Stock Price Trend - {year}-{month:02d}')
            plt.xlabel('Date')
            plt.ylabel('Stock Price')
            plt.legend()
            plt.show()

There is a sharp decline in Pegasus share prices towards the end of February. When we look at the period news, we see that two of THY's planes broke down in similar periods. These malfunctions at THY may have affected the airline industry. (THY plane made a forced landing in the USA: https://www.aa.com.tr/tr/dunya/thy-ucagi-abdye-zorunlu-inis-yapti/757937 & THY plane was evacuated because of the bomb note in the toilet: https://www.sozcu.com.tr/2017/gundem/thy-ucagi-tuvaletteki-bomba-notu-yuzunden-bosaltildi-1687310/) Additionally, we can see a significant increase in the google searches of plane accidents. We can say that these accidents. It may influenced the stock prices since both Pegasus and THY are in the same sector.

Insights for THY Outliers

from IPython.display import display
thyao_data = outliers_thyao[['year', 'month', 'day', 'timestamp', 'THYAO']]
display(thyao_data)

	year	month	day	timestamp	THYAO
31045	2017	2	28	2017-02-28 15:15:00	5.46
31046	2017	2	28	2017-02-28 15:30:00	5.45
31047	2017	2	28	2017-02-28 15:45:00	5.42
31048	2017	2	28	2017-02-28 16:00:00	5.42
31049	2017	2	28	2017-02-28 16:15:00	5.44
31050	2017	2	28	2017-02-28 16:30:00	5.46
31051	2017	2	28	2017-02-28 16:45:00	5.45
31052	2017	2	28	2017-02-28 17:00:00	5.46
31053	2017	2	28	2017-02-28 17:15:00	5.46
31054	2017	2	28	2017-02-28 17:30:00	5.45
31055	2017	2	28	2017-02-28 17:45:00	5.45
31056	2017	2	28	2017-02-28 18:00:00	5.46
42723	2018	8	17	2018-08-17 10:15:00	14.65
42724	2018	8	17	2018-08-17 10:30:00	14.35
42725	2018	8	17	2018-08-17 10:45:00	14.48
42727	2018	8	17	2018-08-17 11:15:00	14.67
42728	2018	8	17	2018-08-17 11:30:00	14.62
42729	2018	8	17	2018-08-17 11:45:00	14.57
42730	2018	8	17	2018-08-17 12:00:00	14.60
42731	2018	8	17	2018-08-17 12:15:00	14.52
42732	2018	8	17	2018-08-17 12:30:00	14.58
42733	2018	8	17	2018-08-17 12:45:00	14.57
42735	2018	8	17	2018-08-17 13:45:00	14.59
47604	2019	4	1	2019-04-01 10:00:00	12.62
47605	2019	4	1	2019-04-01 10:15:00	12.60
47606	2019	4	1	2019-04-01 10:30:00	12.70

# Plotting each company's stock price trend with outliers for each month
company = 'THYAO'
for year in range(2017, 2020):  # Years 2017 to 2019
    for month in range(1, 13):  # Months January to December
        monthly_data = filtered_df[(filtered_df['year'] == year) & (filtered_df['month'] == month)]
            
        if not monthly_data.empty:
            plt.figure(figsize=(10, 5))
            plt.plot(monthly_data['timestamp_2'], monthly_data[company], label=f'{company} Stock Price', linestyle='-')
                
            # Highlight outliers for the specific month
            monthly_outliers = monthly_data[monthly_data[f'{company}_outlier']]
            plt.scatter(monthly_outliers['timestamp_2'], monthly_outliers[company], color='red', label='Outliers', zorder=5)
                
            plt.title(f'{company} Stock Price Trend - {year}-{month:02d}')
            plt.xlabel('Date')
            plt.ylabel('Stock Price')
            plt.legend()
            plt.show()

The decrease in stock prices can be also observed in THY at the end of february like Pegasus. As I mentioned before, two of THY's planes broke down in the second half of February. These malfunctions at THY may have affected the airline industry. (THY plane made a forced landing in the USA: https://www.aa.com.tr/tr/dunya/thy-ucagi-abdye-zorunlu-inis-yapti/757937 & THY plane was evacuated because of the bomb note in the toilet: https://www.sozcu.com.tr/2017/gundem/thy-ucagi-tuvaletteki-bomba-notu-yuzunden-bosaltildi-1687310/) Additionally, THY announced a loss of 6.5 billion lira on February 16. (https://www.turizmguncel.com/haber/thy-65-milyar-lira-zarar-acikladi-chp-arastirma-onergesi-verdi-h30666.html) The effect of this new may additionally reduce stock prices.

There is a significant decrease at August 17, 2018. A campaign not to advertise to US companies was launched, and THY announced its participation in the campaign on August 14. (https://tr.investing.com/news/stock-market-news/abdli-sirketlere-reklam-vermeme-kampanyas-baslatld-thy-ve-turk-telekom-kampanyaya-katldgn-duyurdu-531267) It was stated that with the campaign, organizations that advertise on FOX TV, Google, Facebook, YouTube, Twitter and Instagram were warned and the advertisers were given the message "Stay with Turkey in the economic war". With this statement, the change in marketing strategy at THY may have reduced stock prices.

Insights for VESTEL Outliers

from IPython.display import display
vestl_data = outliers_vestl[['year', 'month', 'day', 'timestamp', 'VESTL']]
display(vestl_data)

	year	month	day	timestamp	VESTL
30208	2017	1	20	2017-01-20 16:30:00	7.05
30209	2017	1	20	2017-01-20 16:45:00	7.05
30210	2017	1	20	2017-01-20 17:00:00	7.04
30212	2017	1	20	2017-01-20 17:30:00	7.03
30213	2017	1	20	2017-01-20 17:45:00	7.12
30214	2017	1	20	2017-01-20 18:00:00	7.09
30215	2017	1	23	2017-01-23 09:45:00	7.13
30216	2017	1	23	2017-01-23 10:00:00	7.14
30217	2017	1	23	2017-01-23 10:15:00	7.03
30218	2017	1	23	2017-01-23 10:30:00	7.08
30219	2017	1	23	2017-01-23 10:45:00	7.10
32419	2017	5	2	2017-05-02 11:45:00	7.39
32420	2017	5	2	2017-05-02 12:00:00	7.41
32421	2017	5	2	2017-05-02 12:15:00	7.40
32422	2017	5	2	2017-05-02 12:30:00	7.41
32423	2017	5	2	2017-05-02 12:45:00	7.41
32425	2017	5	2	2017-05-02 13:45:00	7.41
32426	2017	5	2	2017-05-02 14:00:00	7.42
32427	2017	5	2	2017-05-02 14:15:00	7.40
32431	2017	5	2	2017-05-02 15:15:00	7.39
36313	2017	10	31	2017-10-31 14:30:00	9.15
36314	2017	10	31	2017-10-31 14:45:00	9.12
36315	2017	10	31	2017-10-31 15:00:00	9.10
36316	2017	10	31	2017-10-31 15:15:00	9.08
36317	2017	10	31	2017-10-31 15:30:00	9.10
36318	2017	10	31	2017-10-31 15:45:00	9.21
36319	2017	10	31	2017-10-31 16:00:00	9.38
36320	2017	10	31	2017-10-31 16:15:00	9.40
36321	2017	10	31	2017-10-31 16:30:00	9.65
36322	2017	10	31	2017-10-31 16:45:00	9.86
36323	2017	10	31	2017-10-31 17:00:00	9.92
36324	2017	10	31	2017-10-31 17:15:00	9.96
36325	2017	10	31	2017-10-31 17:30:00	9.90
36326	2017	10	31	2017-10-31 17:45:00	9.86
36327	2017	10	31	2017-10-31 18:00:00	9.88
36338	2017	11	1	2017-11-01 12:15:00	11.68
36343	2017	11	1	2017-11-01 14:00:00	11.61
37685	2018	1	2	2018-01-02 09:45:00	7.61
37686	2018	1	2	2018-01-02 10:00:00	7.64
37687	2018	1	2	2018-01-02 10:15:00	7.81
37688	2018	1	2	2018-01-02 10:30:00	7.86
37689	2018	1	2	2018-01-02 10:45:00	7.83
37690	2018	1	2	2018-01-02 11:00:00	7.88
38923	2018	2	26	2018-02-26 11:30:00	11.78
38924	2018	2	26	2018-02-26 11:45:00	11.69
38925	2018	2	26	2018-02-26 12:00:00	11.67
38926	2018	2	26	2018-02-26 12:15:00	11.65
38927	2018	2	26	2018-02-26 12:30:00	11.65
38928	2018	2	26	2018-02-26 12:45:00	11.68
38929	2018	2	26	2018-02-26 13:45:00	11.68
38930	2018	2	26	2018-02-26 14:00:00	11.83
38931	2018	2	26	2018-02-26 14:15:00	11.73
38932	2018	2	26	2018-02-26 14:30:00	11.62
38944	2018	2	26	2018-02-26 17:30:00	11.66
38945	2018	2	26	2018-02-26 17:45:00	11.75
38946	2018	2	26	2018-02-26 18:00:00	11.77
38947	2018	2	27	2018-02-27 09:45:00	11.67
46902	2019	2	28	2019-02-28 10:30:00	11.08
46903	2019	2	28	2019-02-28 10:45:00	10.85

# Plotting each company's stock price trend with outliers for each month
company = 'VESTL'
for year in range(2017, 2020):  # Years 2017 to 2019
    for month in range(1, 13):  # Months January to December
        monthly_data = filtered_df[(filtered_df['year'] == year) & (filtered_df['month'] == month)]
            
        if not monthly_data.empty:
            plt.figure(figsize=(10, 5))
            plt.plot(monthly_data['timestamp_2'], monthly_data[company], label=f'{company} Stock Price', linestyle='-')
                
            # Highlight outliers for the specific month
            monthly_outliers = monthly_data[monthly_data[f'{company}_outlier']]
            plt.scatter(monthly_outliers['timestamp_2'], monthly_outliers[company], color='red', label='Outliers', zorder=5)
                
            plt.title(f'{company} Stock Price Trend - {year}-{month:02d}')
            plt.xlabel('Date')
            plt.ylabel('Stock Price')
            plt.legend()
            plt.show()

There is an increase in stock prices on 20 January 2017. There is a slight increase on Google Trends data as well. Also, it is said that Vestel Elektronik, traded under BIST 100, showed the best performance of the session in some news. (https://tr.investing.com/news/stock-market-news/turkiye-piyasalar-kapansta-yukseldi-bist-100-093-deger-kazand-123587)

There is not a significant correlation between stock prices and Google Trends for the outlier on 2 May 2017. There is some news indicating that The championship match of the volleyball league organized by Vestel Venus was held on that day. (https://www.aa.com.tr/tr/gunun-basliklari/filede-sampiyon-fenerbahce/809899#) This may be a possible reason for the increase.

At the end of October and beginning of November 2017, companies that will produce domestic automobiles have been announced, including Vestel. (https://www.cnnturk.com/otomobil/yerli-otomobili-uretecek-firmalar-belli-oldu) This seems like strong evidence behind the increase on that day. This can also be seen from Google Trends search keywords “vestel hisse” and “yerli araba”.

There are several outliers on 26, 27 and 28 February 2018. This also aligns with Google Trends. One possible reason for these peaks is that it is announced on the same day that Zorlu Holding has partnered with Chinese GSR, and with this partnership, it will start producing batteries to be used in electric vehicles with an investment of 4.5 billion dollars. (https://www.sozcu.com.tr/2018/ekonomi/zorlu-holding-cinli-gsr-ile-batarya-uretecek-2247943/)